Across the energy value chain, a new breed of domain software applications is rapidly becoming available. These applications are cloud-native and adaptable, with new functionality continuously deployed.
For real-time production optimization, we need novel methodologies that can integrate traditional physics-based modeling with machine learning techniques. Such true hybrid models are powerful. By combining the strengths of both approaches, new and fully automated real-time systems can deliver more accurate, scalable and cost-efficient insights. The outcome – significantly improved efficiency of hydrocarbon production for operators! And as we know, efficiency is what our industry demands in the years to come.
The Window of Opportunity for New Software
Let’s take a step back and review how the window of opportunity for new software has evolved. Over the last decade, the market environment for operators has changed dramatically. Initially, we saw a focus on production growth in an environment where the world’s demand for hydrocarbons grew steadily and oil prices were high. It then turned to a focus on production efficiencies and cost reductions in a more challenging market environment and lower oil prices. Lately, the focus on sustainable operations and carbon-effective production has become a firm requirement and key driver for operations and investments.
Equally so, over the last decade the IT industry has made giant leaps forward. With the digitalization wave and the rapidly increasing amount of data which has become available to the operators, cloud-based thinking has become the new norm – also in our industry. As with previous step-changes in IT, this means that legacy software tools developed for outdated and localized IT infrastructure will find it difficult to keep up on innovation and in transitioning.
The rapidly changing oil & gas market environment combined with the digitalization wave opens a window of opportunity for a new breed of cloud-native domain software. Such software solutions are developed and optimized for reaping the benefits of cloud environments in relation to distribution, scalability and elasticity. Emphasis must be put on novel software for sustainable practices and for production optimization.
The New Generation of Domain Software for Production Optimization
Within the production domain, operators have made significant investments into IIoT, including connected sensors from oil & gas production systems. This data is now made available for flexible consumption from cloud-based data platforms. Having the data infrastructure in place, the operators’ focus turns towards understanding how data is best utilized in order to create value. The industry needs to move beyond bespoke POCs, which rarely make it into regular operations. It is now about embracing commercial software, feeding off these data platforms, at scale.
What should you expect from such cloud-native software solutions? They should be:
- cloud platform agnostic
- extremely fast to deploy with a high degree of automation
- run in real-time and utilize data to the fullest
- a living asset model – an always up-to-date representation of the asset
- a part of open technology, connectible through public API’s
Hybrid Modeling Assures Unmatched Capabilities
There are two distinct approaches which are often set against each other – first-principle physics and machine learning.
In Turbulent Flux we acknowledge that both approaches have their strengths and weaknesses. We know the optimal approach is a combination of the two – a hybrid model of the production system. For us it is a prerequisite to utilize the first-principle physics available from the production system at the core. They provide both predictive and extrapolating capabilities. In addition, machine learning has significant benefits when it comes to scalability, speed and data input flexibility. In our solutions the two approaches feed off each other to leverage their respective strengths. Consequently, this hybrid approach offers unmatched capabilities that only the combination of the approaches can assure.
So let us look at a concrete use case where our FLUX Virtual Flow Meter is used to deliver real-time well rates for gas, oil and water. A pure physics-based solution can deliver very accurate measurements for gas and total liquid. The best possible accuracy for the 3-phase well rates are attained when combined with a machine learning model. This helps to accurately predict the water and oil ratio. Hence, the solution is a hybrid model orchestrated in one system, which accurately predicts all three phases irrespective of constantly changing operating conditions.
Know Your Well Rates!
For effective production optimization, any production system requires an understanding of the productivity of each step in the production process. An oil & gas production system is not an exception. This is why real-time access to multiphase flow rates from all wells and pipelines on an asset is of critical importance. With this information, operators can react immediately. Even more importantly, they can pro-actively optimize operations and avoid deferred or even lost production. A trial-and-error approach to production optimization is no longer satisfactory.
For the best possible outcome, it is essential that one can address the full operating envelope throughout the life-cycle of an asset. To deliver the most accurate live well rates for effective production optimization, we in Turbulent Flux are convinced the best approach is through hybrid, transient (process) modeling.
Turbulent Flux helps you to optimize your production
With Turbulent Flux approach to model-based software, we have responded to our industry’s need for innovation. Legacy software for production operations which runs offline and manually consumes data is a thing of the past. We now require our domain software to fully leverage all the benefits of a modern cloud-based IT and data infrastructure. The basis for all effective production optimization is real-time access to all well rates in a production system. This is what we provide at Turbulent Flux.